[CF-metadata] "positive" attribute

From: Bryan Lawrence <b.n.lawrence>
Date: Thu, 28 Jun 2007 15:56:49 +0100

Hi Forrest

I fear that already we have some default practise in our standard names that
isn't "best" practise ...

... it's my opinion that it would be significantly better to get this right in
the future rather than muddling on with the "way we used to do it". I'll shut
up now and let Jonathan, Roy and Alison wade in [Since my definition of
right may not be :-)]

Cheers
Bryan


On Thursday 28 June 2007 15:07:38 Forrest Hoffman wrote:
> Bryan:
>
> That's what made sense to me as well, but it appears that directionality
> is included in many standard_names already whenever that sign convention
> is not implied by a process-based description in the standard_name.
>
> It would be helpful if a consensus were reached on this issue before I
> start proposing lots of new standard_names.
>
> Forrest
>
> Bryan Lawrence wrote:
> > Your original email started me worrying about a number of things. I'm
> > aware that we're busy building standard names with embedded sign
> > conventions, and I'm not entirely comfortable about that.
> >
> > What happens if I build (or have) a model which happens to have a
> > different sign convention than the one chosen thus far? Do I have to
> > reprocess all my data, or introduce a duplicate set of standard names
> > with the opposite convention embedded. The latter doesn't seem right!
> >
> > I would rather we expanded the use of the positive attribute in a
> > sensible manner in conjunction with new groupings of standard names.
> >
> > Bryan
> >
> > On Thursday 28 June 2007 14:45:11 Forrest Hoffman wrote:
> >> Bryan:
> >>
> >> Jonathan Gregory has been offering suggestions and help off list for a
> >> suite of new standard_names I would like to propose. Jonathan pointed
> >> out that the sign convention is specified in the standard_name, so use
> >> of the "positive" attribute is not needed.
> >>
> >> Thanks!
> >>
> >> Forrest
> >>
> >> Bryan Lawrence wrote:
> >>> Hi Forrest
> >>>
> >>> Are some of these going to be implicit in the standard name? (I imagine
> >>> no always, so I can see the sense of the argument). Personally I don't
> >>> have a problem with introducing an extended controlled vocabulary into
> >>> the positive argument.
> >>>
> >>> Cheers
> >>> Bryan
> >>>
> >>> On Tuesday 26 June 2007 15:03:31 Forrest Hoffman wrote:
> >>>> Much of the CF-1.0 convention seems to be atmospherically biased. I
> >>>> know that's because it is the atmospheric scientists that pioneered
> >>>> all these developments. In attempting to adopt this methodology for
> >>>> other components of Earth System Models, I'm concluding it may be
> >>>> necessary to extend the convention to adequately support the needs of
> >>>> these components.
> >>>>
> >>>> For example, it appears to me that the "positive" attribute for CF-1.0
> >>>> variables is really used to describe vertical directionality. This
> >>>> attribute is often traditionally associated with atmospheric radiation
> >>>> variables. In CMOR, the only valid attribute values are "up" and
> >>>> "down."
> >>>>
> >>>> In terrestrial biogeochemistry various sign conventions exist, and I
> >>>> had hoped to use the "positive" attribute to describe the desired
> >>>> convention where vertical directionality is not important or known.
> >>>> The values I had initially hoped to use for the "positive" attribute
> >>>> for some biogeochemistry variables were "into biosphere" and "out of
> >>>> biosphere" and "into pool" and "out of pool."
> >>>>
> >>>> It would seem that "vertical-directionality" would be a more
> >>>> appropriate attribute where the "positive" attribute has been
> >>>> traditionally applied. I'm sure nobody wants to change this now, but
> >>>> I'm looking for suggestions for attributes that might describe sign
> >>>> conventions where directionality is either not vertical or is more
> >>>> conceptually abstract. We could establish a "negative" attribute, but
> >>>> users might expect some kind of correspondence with the "positive"
> >>>> attribute. A
> >>>> "sign-convention" attribute might work, but it might have many
> >>>> possible values. Any other suggestions?
> >>>>
> >>>> Forrest
Received on Thu Jun 28 2007 - 08:56:49 BST